The proposed research is designed to investigate the use of acoustic spectral information by normal-hearing and hearing-impaired listeners for consonant recognition. By the manipulation of stimulus level and bandwidth parameters of both real and synthetic speech stimuli, the specific critical frequency regions will be defined which are necessary for the perception of certain consonants. The relative importance of certain spetral regions, coupled with their audibility under various listening conditions, may yield a predictive scheme for determining the magnitude of the consonant recognition deficit among individuals with sensorineural hearing loss. Since large differences in speech recognition performance are often observed even with a group of hearing-impaired listeners considered homogeneous on the basis of age, magnitude of loss, etiology, and anatomical site of lesion, a unique approach to subject categorization has been devised. The measurement is based on the refined measurement of selected auditory analysis capabilities which are most likely involved in speech recognition. These measurements have the potential to differentially categorize individuals with hearing loss using criteria which are more relevant to speech recognition processes. The results of these experiments will provide a clearer understanding of the dynamics of the problems in speech recognition and frequency and temporal analysis imposed by sensorineural hearing loss. Practically, the results may yield a method for determing the magnituyde of speech recognition impairment faced by hearing-impared individuals. If subjects with hearing loss can be more effectively evaluated in terms of their receptive communication abilities, the design of amplification systems and improved rehabilitation techniques will be pratical consequences of these results.